Perceived Dementia Risk and Advance Care Planning among Older Adults

Abstract

Background:

Although advance care planning (ACP) is beneficial if dementia develops, and virtually all older adults are at risk for this disease, older adults do not consistently engage in ACP. Health behavior models have highlighted the importance of perceived susceptibility to medical conditions in motivating behavior. Following these models, we sought to determine how often older adults believe they are not at risk of developing dementia and to examine the association between perceived dementia risk and ACP participation.

Methods:

We performed a cross-sectional study of community-dwelling adults without cognitive impairment, aged ≥65, who were interviewed for the Health and Retirement Study in 2016 and asked about their perceived dementia risk (n=711). Perceived dementia risk was ascertained with this question: “on a scale of 0 to 100, what is the percent chance that you will develop dementia sometime in the future?” We used multivariable-adjusted logistic regression to evaluate the association between perceived risk (0% versus >0%) and completion of a living will, appointment of a durable power of attorney for healthcare decisions, and discussion of treatment preferences.

Results:

Among respondents, 10.5% reported a perceived dementia risk of 0%. Perceived risk of 0% was associated with lower odds of completing a living will (OR 0.53; 95% CI, 0.30–0.93) and discussing treatment preferences (OR 0.51; 95% CI, 0.28–0.93), but not appointment of a durable power of attorney (OR 0.77; 95% CI, 0.42–1.39). Many respondents with perceived dementia risk >0% had not completed ACP activities, including a substantial minority of those with perceived risk >50%.

Conclusions:

Older adults with no perceived dementia risk are less likely to participate in several forms of ACP, but the fact that many older adults with high levels of perceived risk had not completed ACP activities suggests that efforts beyond raising risk awareness are needed to increase engagement.

Introduction

The number of Americans with dementia will reach more than 8 million by 2030.¹ Unlike other common diseases, dementia leads to prolonged periods of incapacity during which patients must rely on a surrogate decision-maker. Advance care planning (ACP) is particularly important in this setting. ACP refers to the process by which individuals help to ensure that their preferences inform future treatment decisions.² Its components include completion of written advance directives and communication about goals and wishes with loved ones. For patients with dementia, ACP is associated with lower rates of high-intensity end-of-life care, including potentially burdensome hospital transitions,^{3, 4} and reduced stress among family members.⁵ Despite these benefits, ACP is not consistently practiced.⁶

Efforts to improve participation in ACP increasingly use health behavior models to understand why some individuals engage in ACP and others do not.^{7, 8} A key component of these models is the concept of perceived risk. This is the idea that people must believe they are susceptible to a medical problem before they will engage in behaviors that mitigate or protect against adverse outcomes that result from it.⁹ Dementia is a condition for which nearly all older adults are at risk. The incidence of disease increases exponentially with age and affects more than 35% of adults over age 90.¹⁰ It is not known, however, whether older individuals appreciate that they are at risk of developing dementia and whether acknowledgment of that risk is associated with engagement in ACP.

In the present study, we used data from a nationally representative sample of older Americans to examine these issues. We first sought to determine the percentage of individuals who felt they had no risk of developing dementia. We then assessed the association between perceived dementia risk and ACP participation, with the hypothesis that individuals with no perceived risk would be less likely to have engaged in ACP than those with some perceived risk.

Methods

Data Sources and Study Sample

Respondents were interviewed for the Health and Retirement Study (HRS), which involves a nationally representative cohort of persons over age 50 who are surveyed biennially.¹¹ In 2016, a module containing questions about perceived dementia risk was administered to a randomly selected subset of the study sample. Selected respondents were community-dwelling, could not have a diagnosis of Alzheimer’s disease or another dementia, and had to be able to complete the interview without assistance from a proxy. Among 20,912 respondents, 1,655 were assigned the module. We excluded respondents under age 65 (n=842) because questions about ACP are not administered to them. We also excluded respondents who did not complete questions about perceived dementia risk (n=44) and those who had a self-reported dementia diagnosis or were identified as having probable dementia using the Langa-Weir approach (n=58).¹² The final study sample consisted of 711 respondents (Supplementary Figure S1). The research plan was exempt from review by the institutional review board at Yale University because it involved secondary analysis of publicly available data.

Perceived Dementia Risk

The independent variable of interest was perceived dementia risk, measured using this question: “Dementia is a general term for a decline in mental ability severe enough to interfere with daily life. Memory loss is an example. Alzheimer’s is the most common type of dementia. On a scale of 0 to 100, what is the percent chance that you will develop dementia sometime in the future?” After answering, respondents were asked if they had given a rounded or approximate number. If so, they could revise their estimate. Those who gave only a closed range of values, not an integer, were assigned the mean value of the closed range. For example, a respondent who stated that she had a 30%–50% chance of developing dementia was assigned a perceived risk of 40%. Those who gave only an open range were assigned the top or bottom value. For example, if a respondent stated that her risk was 30%, at most, she was assigned a perceived risk of 30%; if a respondent stated that her risk was 80%, or more, she was assigned a perceived risk of 80%. No respondents gave an open range with a bottom value of 0%.

We focused on respondents who felt they had no dementia risk because this view is almost always incorrect. While a wide range of estimates are potentially accurate because of the heterogeneity in dementia risk factors among older adults, a risk estimate of 0% is too low since virtually all older persons have some risk of developing dementia in their lifetimes.¹³ For this reason, we treated perceived dementia risk as a binary variable (0% or >0%). We also performed sensitivity analyses in which we treated perceived dementia risk as a categorical variable, divided in ten percent increments (0%, 1–10%, 11–20%, 21–30%, 31–40%, 41–50%, and >50%).

Advance Care Planning

Our primary outcomes were three ACP activities that are discussed in another section of the HRS survey. Since 2012, respondents aged 65 and older have been asked: (1) if they have appointed a durable power of attorney for healthcare decisions; (2) if they have completed a living will; and (3) if they have “ever discussed with anyone the care or medical treatment [they] would want to receive if [they] were to become seriously ill in the future.” We considered respondents to have discussed their treatment preferences if they answered “yes” to this question in 2012, 2014, or 2016. We considered respondents to have a durable power of attorney or living will if they confirmed these arrangements were in place in 2016.

Covariates

We selected covariates for inclusion a priori after reviewing the existing literature on factors associated with ACP participation.^14–16 These included age, gender, marital status, race and ethnicity, highest level of educational attainment, household wealth, self-rated health, number of chronic conditions, and depression.¹⁷

Statistical Analysis

We described the demographic and health characteristics of the sample and used separate logistic regression models to examine the association between no perceived dementia risk (versus some perceived risk) and each ACP behavior, adjusting for the covariates above. We repeated the multivariable modeling in sensitivity analyses treating perceived dementia risk as a categorical variable. Analyses incorporated the survey design variables and weights to account for differential nonresponse and the complex sampling design,¹⁸ and were performed using Stata 17 (StataCorp, College Station, TX); a p-value <0.05 denoted statistical significance.

Results

The characteristics of the study sample appear in the Table. The largest percentage of respondents were between 70 and 79 years old. A majority were female and White, nearly three quarters rated their health as “good” or better, and most had at least two chronic conditions.

Table.

Characteristics of the study sample.

Characteristic – Weighted %a	Respondents (n = 711)
Age, years
65–69	35.0
70–79	44.4
≥80	20.5
Female	61.1
Married or living with a partner	56.6
Race and Ethnicity
White, non-Hispanic	83.5
Black, non-Hispanic	8.0
Hispanic	5.7
Other	2.8
Educational level
Less than high school	12.0
High school	32.3
Some college	28.6
College or more	27.2
Household wealth
<$15,000	15.4
$15,000 – $199,000	35.4
$200,000 +	49.2
Self-rated health
Poor or fair	26.1
Good	35.6
Very good or excellent	38.3
Total number of chronic conditionsb
0–1	26.1
2–3	51.2
4+	22.7
8-item CES-Dc score
0–2	81.2
3+	18.8

^aPercentages were derived using the HRS sampling weights and may not sum to 100% because of rounding. Data on marital status were missing for 1 respondent.

^bHigh blood pressure, lung disease, arthritis, heart disease, stroke, psychiatric problems, diabetes, and cancer.

^cCES-D = Center for Epidemiologic Studies Depression Scale. Scores range from 0 to 8, with higher scores indicating more depressive symptoms.

The distribution of perceived dementia risk is shown in Figure 1. Among respondents, 10.5% had a perceived dementia risk of 0%. By contrast, 89.5% of respondents had a perceived risk >0%, with 14.7% reporting a perceived risk >50%.

Figure 1.

Distribution of perceived dementia risk, as measured using this question: “on a scale of 0 to 100, what is the percent chance that you will develop dementia sometime in the future?” Respondent percentages were derived using the HRS sampling weights.

Rates of engagement in each ACP activity, stratified by perceived dementia risk, are shown in Figure 2. Among respondents with 0% perceived risk, 40.6% had completed a living will, 62.4% had discussed their treatment preferences, and 48.6% had appointed a durable power of attorney. Among all respondents with >0% perceived risk, 59.8% had completed a living will, 79.9% had discussed their treatment preferences, and 60.2% had appointed a durable power of attorney. For those with perceived risk >50%, 60.5% had completed a living will, 84.4% had discussed their treatment preferences, and 59.2% had appointed a durable power of attorney.

Figure 2.

Relationship between perceived dementia risk and advance care planning (ACP) activities. The horizontal bars indicate pooled percentages among all respondents with perceived risk >0%. An asterisk denotes a statistically significant difference between this group and those respondents with perceived risk of 0%. Data about ACP participation were missing for 1 respondent.

In multivariable-adjusted logistic regression models, perceived dementia risk of 0% was associated with lower odds of completing a living will (OR 0.53 [95% CI, 0.30–0.93]) and lower odds of discussing healthcare preferences (OR 0.51 [95% CI, 0.28–0.93]), compared to perceived dementia risk >0%. There was not a significant association between perceived dementia risk and appointment of a durable power of attorney (OR 0.77 [95% CI, 0.42–1.39]). Full model results appear in Supplementary Tables S1–S3. Sensitivity analyses treating perceived dementia risk as a categorical variable showed similar results (Supplementary Table S4).

Discussion

In a nationally representative sample of community-dwelling older Americans without cognitive impairment, we examined the relationship between perceived dementia risk and ACP engagement. We found that more than 1 in 10 older adults believed they had no chance of developing dementia and that such persons had significantly lower rates of completing a living will and discussing their treatment preferences with someone else. They were also less likely to have appointed a durable power of attorney for healthcare decisions, but this association was not statistically significant. While participation in ACP was higher among individuals with some perceived dementia risk, many individuals — even those with high perceived risk — had not engaged in these activities. For example, approximately 40% of persons with a perceived dementia risk >50% had not completed a living will or appointed a durable power of attorney.

Awareness of risk has a central role in theories of health behavior.⁹ It has been associated with participation in health-promoting activities across multiple domains, including cancer screening,¹⁹ smoking cessation,²⁰ and vaccine use.²¹ Our results extend this work by showing that a small but meaningful number of older adults do not believe they are at risk for dementia and that such persons are much less likely to have engaged in ACP. Encouraging patients to share how they view their chances of developing dementia may provide an opportunity, for those who believe they have no risk, to correct that misapprehension. Better awareness of dementia risk may help some patients to look ahead to a time of decisional incapacity and may serve as a motivation for ACP participation.

A substantial minority of individuals who perceived they were at risk for dementia had not completed ACP activities, however. This was true even among respondents who believed they were more likely than not to develop this disease. This observation calls attention not just to the diverse set of factors that influence participation in ACP^22–24 but also to the importance of talking explicitly to patients about ACP in the context of dementia risk and explaining why ACP will be beneficial if dementia develops. The available evidence suggests that there are persons who acknowledge their risk of developing dementia but do not feel ACP is relevant to them.^{7, 25} Many older adults and their families do not understand the natural history of dementia and fail to recognize it as a life-limiting illness.²⁶ There is also a broader need for ACP models that speak, in a more compelling way, to the unique challenges that dementia presents.²⁷ Patients are often encouraged to think about ACP in terms of decisions about life-sustaining treatment during a period of incapacity that is assumed to be abrupt and brief. This conceptualization of ACP is out of step with the trajectory of dementia, which leads to incapacity that is insidious and prolonged.²⁵

This study has several limitations. Because questions about perceived dementia risk were asked in a single HRS module, we could not examine whether risk perceptions changed over time. The HRS survey does not routinely collect risk estimates for other diseases; although our models were adjusted for self-rated health and the presence of chronic conditions, there may be residual confounding from respondents’ perceived risk of medical problems besides dementia. We did not have information about why respondents did or did not engage in ACP. Finally, perceived risk is influenced by anchoring and availability heuristics,²⁸ issues with numeracy,²⁹ and a tendency towards qualitative, “gist” estimates rather than quantitative ones.³⁰ For this reason, we used the simplest possible construct (no risk versus some risk) and performed sensitivity analyses in which our results were similar, even when perceived risk was operationalized differently.

In summary, we found that more than 1 in 10 older persons believed they had no chance of developing dementia and that such persons were much less likely to have engaged in several ACP activities. While it is important for healthcare professionals to address this misperception, the fact that substantial numbers of persons with high perceived risk had not completed ACP suggests that improved risk awareness alone will not lead many individuals to participate.

Supplementary Material

supinfo

Supplementary Figure S1. Construction of the study sample.

Supplementary Table S1. Full results of logistic regression model examining the association between perceived dementia risk and completion of a living will.

Supplementary Table S2. Full results of logistic regression model examining the association between perceived dementia risk and discussion of healthcare treatment preferences.

Supplementary Table S3. Full results of logistic regression model examining the association between perceived dementia risk and appointment of a durable power of attorney for healthcare decisions.

Supplementary Table S4. Sensitivity analyses examining the association between perceived dementia risk and advance care planning activities, with perceived dementia risk operationalized as a categorical variable.

Key points:

• When community-dwelling older adults were asked to estimate their risk of developing dementia, more than 10% felt they had no risk.

• Those with no perceived dementia risk were much less likely to have completed several advance care planning activities.

• Participation in advance care planning was far from universal even among older adults with very high perceived dementia risk.

Why does this paper matter?

Virtually all older adults are at risk for dementia, and raising awareness of that risk may help motivate engagement in advance care planning. For many older adults, however, increasing risk awareness alone is unlikely to be enough to lead them to participate.